A data-driven procedure to model occupancy and occupant-related electric load profiles in residential buildings for energy simulation

Improving the reliability of energy simulation outputs is becoming a pressing task to reduce the performance gap between the design and the operation of buildings. Occupant behaviour modelling is one of the most relevant sources of uncertainty in building energy modelling and is typically modelled via a priori choices made by modellers. Thus, an improvement in the description of occupant behaviour is needed. To this regard, the availability of smart meter recordings might help to generate more reliable input data for building energy models. This paper discusses a novel data-driven procedure that enables to create yearly occupancy and occupant-related electric load profiles to inform building energy modelling, using a typical uneven database made available by energy operators. The procedure is subdivided into three main tasks. The first has the intent to detect representative occupant-related electric load profiles from smart meters readings. The second task aims to generate yearly occupancy profiles from the same database. The last task assesses the impact of the generated occupancy and occupant-related electric load profiles on building energy simulation outputs. The procedure is applied to the case study of a multi-residential building in Milan, Italy and is meant to show the possibility to overcome deterministic inputs that might have little relation with the actual building operation. It showed a substantial improvement in the reliability of building energy simulation and that occupant related load profiles may account for about 8% of the building's energy need for space heating.